def language(self):
if self.hint_language_code is not None:
self.__lang = Language.from_code(self.hint_language_code)
if self.__lang is None:
self.__lang = self.detected_languages.language
return self.__lang
def __init__(self, text, lang_code=None, word_tokenizer=None,
sentiment_weighting=None):
if not isinstance(text, basestring):
raise TypeError('The `text` argument passed to `__init__(text)` '
'must be a unicode string, not {0}'.format(type(text)))
self.raw = text
if not isinstance(text, unicode):
self.raw = text.decode("utf-8")
self.string = self.raw
if lang_code is not None:
self.__lang = Language.from_code(lang_code)
else:
self.__lang = self.detected_languages.language
if word_tokenizer is not None:
self.__word_tokenizer = word_tokenizer
else:
self.__word_tokenizer = WordTokenizer(locale=self.language.code)
if sentiment_weighting is not None:
self.__sentiment_weighting = sentiment_weighting
else:
self.__sentiment_weighting = long_text_weighting
def language(self):
if self.hint_language_code is not None:
self.__lang = Language.from_code(self.hint_language_code)
if self.__lang is None:
self.__lang = self.detected_languages.language
return self.__lang
def language(self, value):
self.__lang = Language.from_code(value)
def language(self, value): self.__lang = Language.from_code(value)
def polyglot_default_install():
"""checking and caching polyglot"""
try:
os.path.sep = '/'
polyglot_path = POLIGLOT['path_polyglot_data']
polyglot_path = tools.get_abs_path(polyglot_path)
if not os.path.exists(polyglot_path):
os.makedirs(polyglot_path)
load.polyglot_path = polyglot_path
downloader = Downloader(download_dir=polyglot_path)
for language in SERVER['language']:
sentiment = False
entities = False
morph = False
pos = False
full_name = Language.from_code(language).name
if language in get_supported_languages('embeddings2'):
#if not downloader.is_installed(unicode('embeddings2.' + language)):
if not downloader.is_installed('embeddings2.' + language):
raise EnvironmentError(
'The {0} module for {1}({2}) was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'embeddings2', full_name, language))
if language in get_supported_languages('ner2'):
entities = True
#if not downloader.is_installed(unicode('ner2.' + language)):
if not downloader.is_installed('ner2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'ner2', full_name))
if language in get_supported_languages('sentiment2'):
sentiment = True
#if not downloader.is_installed(unicode('sentiment2.' + language)):
if not downloader.is_installed('sentiment2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'sentiment2', full_name))
if language in get_supported_languages('morph2'):
morph = True
# if not downloader.is_installed(unicode('morph2.' + language)):
if not downloader.is_installed('morph2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'morph2', full_name))
if language in get_supported_languages('pos2'):
pos = True
# if not downloader.is_installed(unicode('pos2.' + language)):
if not downloader.is_installed('pos2.' + language):
raise EnvironmentError(
'The {0} module for {1} was not found, to install this package,'
' run "./install/install_polyglot.py"'.format(
'pos2', full_name))
# FOR POLYGLOT DOWNLOAD ON START SERVER
if sentiment or entities or morph:
text_polyglot = Text('Testing and cashing',
hint_language_code=language)
if sentiment:
_ = text_polyglot.words[0].polarity
_ = text_polyglot.sentences[0].polarity
if entities:
_ = text_polyglot.entities
if morph:
_ = text_polyglot.morphemes
except:
ex_type, ex, tb = sys.exc_info()
# tools.message_box(str(ex) + 'TracebackError'+''.join(traceback.format_exc()),
# str(ex_type), 0)
raise EnvironmentError(
str(ex) + 'TracebackError' + ''.join(traceback.format_exc()))
